Skip to content

TheGoat395/Codex-Skills / skills/deep-research

deep-research

Verified

Research with scoped depth and counterevidence.

references/coverage-and-research-passes.md

View on GitHub
# Coverage And Research Passes

Loaded only for the mode declared in [SKILL.md](../SKILL.md). Apply the relevant sections; preserve the requested objective and scope.

## Derive the Coverage Graph

Before searching, infer the dimensions that could materially change the answer. Represent the task as a dynamic graph containing only applicable elements:

- questions, claims, unknowns, and decision criteria;
- relevant entities, artifacts, events, systems, and alternatives;
- relationships, dependencies, incentives, and conflicts;
- authoritative, independent, experiential, and contradictory evidence classes;
- time, version, geography, population, plan, environment, or other scope boundaries;
- available access surfaces and materially different fallback paths;
- practical constraints such as cost, capacity, complexity, ownership, risk, and scale;
- existing decisions or baselines that the result must confirm, revise, or reject.

Expand the graph when evidence reveals a material adjacent branch. Prune branches that cannot affect the decision. This graph, not a hardcoded source list or quota, determines coverage.

## Run the Research Passes

### 1. Frame

Build the coverage graph, define comparison criteria, expose assumptions, and identify what would reverse the expected conclusion.

### 2. Discover

Map the credible field before selecting favorites. Search across materially different terminology, source origins, adjacent categories, counterexamples, and relationship paths implied by the graph.

### 3. Screen

Use lightweight evidence to rank relevance and evidentiary value. Deduplicate repeated origins, resolve uncertain identities, remove immaterial candidates, and choose a bounded high-value subset for deeper inspection. Size the funnel to the decision rather than a predetermined count.

### 4. Deepen

Inspect the complete evidence needed for each decisive claim. Preserve dates, scope, definitions, units, versions, exceptions, methods, and incentives. Follow citations and dependencies back to their strongest practical origin.

### 5. Challenge

Seek credible contrary evidence, negative cases, alternate explanations, incentive conflicts, methodological weaknesses, implementation failures, and omitted categories. Resolve conflicts by authority, currency, scope, method, and directness; leave genuine disputes visible.

### 6. Synthesize and Verify

Compare alternatives using one consistent framework. Separate verified fact, source-supported interpretation, inference, estimate, recommendation, disputed claim, and unknown. Verify that every material citation supports the exact nearby claim.
add-product-e2e-evalpaperclipai/paperclipAdd or extend a Paperclip full-stack runner E2E workflow, fixture, matcher, or report evidence path for local or Daytona execution.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
add-runner-evalpaperclipai/paperclipAdd or extend a Paperclip Runner protocol evaluation definition, roster, assertion, or report fixture with provenance and narrow validation.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
diagnose-why-work-stoppedpaperclipai/paperclipCompatibility entry point for failed, stalled, looping, or over-recovered Paperclip work. Use fix-failed-paperclip-run for the full diagnosis, regression, repair, real verification, and green-PR workflow; honor an explicitly requested diagnosis-only scope.Stars: 115kUpdated Claude CodeCodexVerifiedInstall
fix-failed-paperclip-runpaperclipai/paperclipDiagnose and fix failed, stuck, slow, or incorrectly blocked Paperclip tasks and runs. Use with a task/run URL or failure report to trace the root cause, prove a regression red then green, check second-order effects, verify the real workflow, and deliver a PR with passing checks.Stars: 115kUpdated Claude CodeCodexVerifiedInstall